Why has Fibre Channel over Ethernet (FCoE) adoption been slower than expected?
There are three main reasons why FCoE adoption has been sluggish. The first is technology immaturity. When the servers, converged network adapters (CNAs), data center bridging (DCB) switches and storage systems come from a single vendor, interoperability usually isn’t an issue. Neither is it an issue for joint ventures, such as the Cisco-EMC-VMware created Virtual Computing Environment (VCE) or FlexPod infrastructure stack that uses NetApp, Cisco and VMware technology. But when multiple independent vendors are in the mix, interoperability issues arise. Right now, it almost feels as if the current state of FCoE is about where Fibre Channel (FC) was in the late 1990s.
The second reason comes down to a lack of perceived value. The technology's selling point is that it provides a lower total cost of ownership (TCO) than FC. It's supposed to do this primarily by converging the TCP/IP LAN with an FC SAN, thereby reducing duplicate fabrics. However, the relatively high cost of new DCB switches, CNAs and storage systems required to implement FCoE can be daunting.
Much less-expensive FC technologies (now at 16 Gbps, which is backwards-compatible with 8 Gbps and 4 Gbps) and iSCSI running over standard 10 Gbps, 40 Gbps, and 16 Gbps LANs, are often considered good enough or better because of their lower TCO. The latency advantages of FCoE over iSCSI aren't compelling and only valuable in a very narrow set of circumstances.
The third reason for FCoE's slow adoption is the server/tech refresh cycle. Few administrators wish to crack open their servers and put in new adapters. That's why each new server comes with new storage adapters. As server technology refreshes pick up, so will FCoE adoption.
This was first published in October 2012